Production surface — everything below is what the live dashboard would show. This fold is for the presenter.

Talk track. Speed of checkout is their boss’s stated KPI — get customers in and out faster; QuikTrip is the competitor he names. Store to store, then district to district — then it becomes competitive. Their rule of thumb: see a queue, break the line — the second-checkout trigger automates it. 61% choose a manned lane over self-checkout; POS never sees that choice. Waiting is an absence POS cannot record. “Left without reaching the counter” is a camera-observed journey outcome; it does not claim whether a purchase occurred. The checkout queue holds cars on the pumps — the one-system story with Forecourt.

Claims. Numbers are illustrative, on Register.md + journey + live-presence machinery. Serving time per customer is serving seconds ÷ customers served, presence-measured — POS transaction time starts at ticket-open and reads smaller. Queue time is measured in the queue zone, outside the serving/waiting states. Never a bare “served” — the count is customers served; the time-state is serving. Second-checkout trigger is queue depth × a closed lane. In-and-out reuses Walk-in’s dwell for customers who reached a serving state; it does not verify a purchase. The 14 min wait tile is the CX mix with response as a sub. The 11 who left before the counter is a funnel step, not a purchase claim. Staffing mix, response histogram, and queue by daypart stay off this surface — they live on the PathCI week view (checkout-week.html). District rollup is a planned grouping — today the product compares stores.

Build readiness. Serving and waiting states are a shipped primitive. Serving time, queue time, slow checkouts, response time, store comparison, and door-to-served are derivable. Not built: manned vs self-checkout split, left-before-counter, second-checkout trigger, district rollup.